Agent Engineer

We are a small team teaching machines to run compliance work end-to-end. Brayns reads how operators actually work — documents, case history, the judgment of senior people — and turns that structure into agents that execute continuously, traceably, at scale.

We’re building what we think will be one of the next unicorns out of Stockholm, and agents are core to how our product works. We’re hiring an Agent Engineer to design, build, and continuously improve them.

This is a product-focused role. You care about whether the agent actually solves the user’s problem reliably — not just whether the demo looks impressive. You think in terms of real workflows, edge cases, failure modes, and the experience the user is left with after the agent runs.

This role is onsite in Stockholm and open only to candidates who already live in Stockholm or are ready to relocate before starting. We will not be considering remote applicants.

• Design and build agents that solve real product problems end-to-end — not just prototypes.
• Decide which LLM is right for each agent, and back those decisions up with experiments and evals rather than vibes.
• Build the surrounding scaffolding agents need to be reliable in production: tool use, error handling, retries, guardrails, observability.
• Iterate on agents based on real usage — measuring what works, finding where they break, and closing the loop.
• Work closely with the rest of the engineering team to integrate agents into the product cleanly.

• Product-focused. You think about the user, the workflow, and the outcome — not just the model. You’d rather ship a slightly less clever agent that actually works than a fancy one that fails 30% of the time.
• Real agentic platform experience. You’ve built agents inside a real product or platform that real users depend on. Hobby projects, weekend hacks, and tutorial demos don’t count for this one — we want someone who has lived with agents in production.
• Experiment-driven model selection. When asked which LLM to use for a given agent, your answer is grounded in evals and experiments you’ve actually run — not in what’s trending or what you used last time.
• Strong Python. Python is our primary language for this work. We’d like a Python expert, but we can be flexible here for the right person if the rest of the profile is a strong fit.